Twenty-two verified open-source projects have publicly rejected AI-generated contributions, spanning code, documentation, translations, and issue reports. This shows that the debate is shifting from “whether AI can be used to write code” to “who should bear the cost of maintenance.”
What this is
Open Source Says No is a directory that catalogs relevant policies. Projects including yt-dlp, Godot, QEMU, and immich have drawn clear boundaries: they do not accept AI-generated feature code, large-scale refactors, or community-facing text without permission. Godot permits mechanical assistance such as autocomplete and find-and-replace, but requires disclosure and reserves the right to close policy-violating submissions and ban accounts.
Here, “vibe coding” means relying primarily on natural-language instructions to have AI generate an entire program or modify code. Different projects take one of three approaches: banning, disclosure, or allowing contributions with accountability. Their common baseline is that contributors must understand the code, complete testing, and explain every change.
Industry view
We believe the Linux Foundation finding that “roughly one-quarter of new pull requests are estimated by maintainers to be low-quality AI content” deserves more attention than disputes over model capabilities alone. AI makes submissions nearly free, but it does not add reviewers at the same rate. Once generation outpaces review, maintainers have to tighten the entry gate.
The pushback against bans has its own valid basis. Detecting AI-assisted code is unreliable, and automatic bans may punish developers who responsibly use autocomplete tools. One-size-fits-all policies may also drive away new contributors. The risks are not limited to “more junk code”; they include good contributions being wrongly blocked and enforcement becoming dependent on platform judgments and community reports.
Moreover, strict restrictions may push training data, documentation, and localization work back to individuals, small teams, or closed-source projects, without necessarily reducing the software ecosystem’s total costs. Bans can protect some maintainers, but they are not a long-term quality strategy.
Impact on regular people
For enterprise IT: Companies need to incorporate AI assistance into code governance, including source disclosure, testing accountability, and submission review. The question cannot be limited to “Can it generate?” Teams must also ask, “Is there someone who truly understands the output and takes responsibility for it?”
For individual careers: AI will continue to accelerate code and documentation drafting, but job value will increasingly center on judgment, validation, and accountability for outcomes. Demonstrated engineering ability still matters more than knowing how to write prompts.
For the consumer market: Open-source software maintenance is becoming more cautious, and collaboration efficiency may decline in some projects. However, if high-quality tools receive steadier human maintenance as a result, users may not experience worse software quality over the long term.